The proposed research will result in a computerized system that uses fractal algorithms for compressing and decompressing digitized images of chest x-rays and analyzing recurrent geometric patterns. Utilizing a combination of general purpose fractal compression with customized fractal algorithms for recurrent geometrical patterns, this approach promises to provide significant improvement over other compression methods. Phase I research focuses on demonstrating algorithms for specific patterns found in the right lung. Phase II extends the class of algorithms for the entire chest x-ray and incorporates them into the system. A catalog of standardized, recurrent patterns will be developed. Fractal algorithms will provide a measure of variance of test images from standard patterns within the catalog. Efficacy of this variance for diagnosing pulmonary nodules will be explored. Development of an efficient compressing system for chest x-rays will significantly impact electronic medical records systems, and transmission for consultative, diagnostic and research purposes. The computer programs for decompressing images will be designed to be independent of the number or type of customized algorithms used in compression and will be readily implemented on PC's and other computer platforms.